In the Loess Plateau, due to the inappropriate vegetation restoration mode, large areas of artificially restored vegetation began to degrade, thus the optimization of vegetation allocation has become an urgent necessity. The main purpose of this study was to identify and evaluate slope runoff and soil loss rates, and to review all of the plot-scale studies in the Chinese Loess Plateau, by meta-analysis. Based on data collected from the runoff plot, the effect of land use on annual runoff and annual soil loss under natural rainfall conditions was analyzed. The optimization of land use in different climatic regions of the Loess Plateau was evaluated. The plot database contained 55 plot measuring sites in the Loess Plateau, which included 461 runoff plots and 535 soil loss plots. Bare soil was found to have the highest average annual runoff (58.57 mm·yr^-1) and annual soil loss (122.06 t·ha^-1·yr^-1). Natural grassland and mixed forest had the lowest annual runoff (< 15 mm·yr^-1) and annual soil loss (< 20 t·ha^-1·yr^-1), exhibiting a better effect of soil and water conservation when the precipitation was < 200 mm and > 600 mm, respectively. When the precipitation was 400 – 600 mm, shrubland showed the lowest mean annual runoff (21.36 mm·yr^-1) and annual soil loss (13.36 t·ha^-1·yr^-1), which conducive to reducing water and sediment. Therefore, shrubland could be selected as the recovery vegetation type in the semi-humid climatic region. Land-use types determined the relationship between annual soil loss and annual runoff with plot length and slope gradient. These results enabled the assessment of the impact of land-use change on water erosion, providing a basis for formulating soil and water conservation management programs.